American Mineralogist, Volume 67, pages 1058-1064, 1982 A further crystal structure refinement of gersdorffite PETER BAYLISS Department of Geology and Geophysics University of Calgary, Alberta T2N IN4, Canada Abstract Thirteen gersdorffite (NiAsS) specimens were analyzed with electron microprobe, powder diffraction, and precession camera techniques. For each specimen, a sample with a well-developed cube form {lOO} was selected. Intensity data were collected from 27 reflections, which are mostly forbidden by cubic space group Pa3, with a single crystal diffractometer. X-ray powder diffraction data indicate two pyrite subgroup specimens, five ullmannite subgroup specimens, and six cobaltite subgroup specimens. A complete intensity data set was collected by single crystal diffractometer from three of these six cobaltite subgroup samples. Least-squares refinement shows different degrees of apparent As-S disorder with one sample significantly ordered (predominantly three twin-related domains) and the other two samples disordered (equal amounts of six twin-related domains). These three crystal structures are similar to those described for cobaltite. They are explained as a sextuplet of orthorhombic (Pca2t) interpenetrating twin-related domains about a 3 twin axis [111]. Although the chemical compositions of the three different pyrite-type crystal structures overlap, there is a tendency for the two pyrite subgroup (true As-S disorder at the atomic level, Pa3) specimens to contain more As, and a tendency for the six cobaltite subgroup (true As-S order at the atomic level, Pca2t) specimens to contain more Co and Fe than the five ullmannite subgroup (true As-S order at the atomic level, P2t3) specimens. These three crystal structures are probably temperature dependent with P2t3 the low form, Pca21 the intermediate (metastable?) form, and Pa3 the high form. Introduction ROM M15861) with Straumanis-type powder photo- graphs, Cu radiation and Ni filter. They state "pat- Three crystal structure variants of gersdorffite tern similar to that of cobaltite", although they have been described as follows: an ordered pyrite- observed the 010 and 110 reflections in cobaltite type structure (P213, ullmannite subgroup, Bayliss from Hakensbo. These weak 010 and llO reflections and Stephenson, 1967); a disordered pyrite-type are difficult to observe, since they may be hidden by structure (Pa3, pyrite subgroup, Bayliss 1968); and the high background fluorescence from the Fe and a distorted disordered pyrite-type structure (Pl, Co in gersdorffite with Cu radiation. These 010 and gersdorffite III, Bayliss and Stephenson, 1968). An 110 reflections were observed in both Debye-Scher- alternative interpretation to these diffraction data rer photographs and powder diffractometer patterns for gersdorffite III is suggested by the work of in six gersdorffite specimens (USNM R830, Bayliss (1982), who interprets different As-S order- BMl922,145, USNM R862, BMI929,12, HMM, disorder within cobaltite crystals as a sextuplet of BM57562) by Bayliss (1969a). orthorhombic (Pca2) interpenetrating twin-related Klemm (1962), Ramdohr (1969), and Cabri and domains about a:3 twin axis [111]. In view of this, Laflamme (1975) indicate that most specimens of the crystal structure of gersdorffite III is re-exam- gersdorffite are optically isotropic, but some are ined in more detail. optically anisotropic. Bayliss (1969a) shows a grad- The 010 and 110 reflections may be used to ual increase from optically isotropic for NiAsS to differentiate between space groups Pca2], P2]3 and the strongest optical anisotropy for CoAsS in the Pa3. Berry and Thompson (1962) did not observe gersdorffite-cobaltite solid solution series. Optical the 010 and 110 reflections in gersdorffite, (Ni,Fe, anisotropy is most frequently recognized by Klemm Co)AsS, from Sudbury, Ontario (ROM M12176 and (1962) through twin lamellae after {III}? and {IOO}. 0003-OO4X/82/091O-1058$02.00 1058 BAYLISS: GERSDORFFITE 1059 Ramdohr (1969) states "all authentic specimens Table 2. Electron microprobe analyses examined by the writer or by Meyer (1926) did not Specimen exhibit lamellar structure", however, Bayliss Number Fe Co Ni S As Sb Total (1969a) observed twinning in gersdorffite from NiAsS 35.4 19.4 45.2 100.0 Leichtenberg (USNM R862) and Lobenstein (BM57562). 57562 0.2 0.2 34.1 17.8 46.7 0.1 99.1 R862 1.8 0.7 32.9 18.8 45.0 0.4 99.6 1929,12 2.3 0.5 32.1 19.2 44.7 0.2 99.0 Experimental and results M12176 5.3 5.4 23.6 18.2 46.6 0.1 99.2 Thirteen gersdorffite specimens were obtained M15861 6.5 9.3 18.8 18.7 46.8 0.3 100.4 1922,145 1.7 17.8 14.8 16.2 50.0 100.5 for this investigation from the British Museum (BM), Royal Ontario Museum (ROM), United 1959,462 0.4 0.3 34.1 18.3 44.5 1.5 99.1 113044 0.9 1.2 33.6 20.0 45.4 101.1 States National Museum (USNM), and University 1434 1.0 1.4 32.6 18.9 43.6 1.7 99.2 of New South Wales (UNSW). The specimens, 120381 0.6 4.2 27.6 11.0 57.5 100.9 which have previously been studied, are BM 1434, 1917,285 5.6 0.1 29.3 17.1 48.7 0.1 100.9 BM 57562, BM 1917,285, BM 1922,145, BM 1929, UNSW 1.3 0.4 32.0 14.2 52.5 0.4 100.8 12, BM 1933,371, BM 1959,462, USNM R862 and 1933,371 3.3 2.2 26.3 10.1 58.0 0.8 100.7 UNSW by Bayliss (1969a), and ROM M12176 and ROM M15861 by Berry and Thompson (1962). Specimen numbers with their localities are listed in showed distinct reflection splitting. Most specimens Table 1, and the same order is also used in Tables 2 of groups (2) and (3) above show sharp reflections and 3. near () = 900, which indicates cubic symmetry. On Powder diffractometer patterns may be divided the other hand, most specimens of group (1) above () into three groups as follows: (1) both the 010 and show broad reflections near = 900, which may be 110 reflections are observed in six gersdorffite spec- interpreted as multiple reflections from a pseudo- imens (57562, R862, 1929,12, M12176, M1586l, and cubic mineral. 1922,145), (2) the 010 reflection is absent although All gersdorffite specimens were chemically ana- the 110 reflection is observed in five gersdorffite lyzed by electron microprobe as previously de- specimens (1959,462, 113044, 1434, 120381, and scribed by Bayliss (1982). Calculation of the stoichi- 1917,285), and (3) reflections 010 and 110 are absent ometry from electron microprobe results in Table 2 in two gersdorffite specimens (1933,371 and based upon a 4MXn structural formula indicates UNSW). The shapes of reflections 010 and 110 are that n varies between 1.97 and 2.03. Since the similar to those of the other reflections recorded. electron microprobe data contain random errors, no The presence of these 010 and 110 reflections have evidcnce is available to indicate deviations from been confirmed by 114.6 mm Debye-Scherrer pow- stoichiometric MX2. The Ni-As-S system de- der photographs. Neither powder diffractometer scnbed by Yund (1962) gives a MX2 formula for patterns nor Debye-Scherrer powder photographs gersdorffite based upon chemical syntheses and a literature survey. A similar conclusion has been Table I. Specimen numbers and their localities deduced by Klemm (1965). Therefore it appears logical to accept a stoichiometric MX2 formula. Specimen Number Locality The gersdorffite specimens (Table 2) show a wide range of substitution of Ni by Co and Fe, which falls BM57562 Lobenstein, Russ, Germany within the solid solution limits of the ternary dia- USNM R862 Leichtenber9, Fichtelgebirge, Germany BM 1929,12 Mitterburg, Salzburg, Austria gram FeAsS-CoAsS-NiAsS at 5000 C of Klemm ROM M12176 Crean Hill, Sudbury, Ont., Canada (1965). Metal chemical zoning is observed in speci- ROM M15861 Garson Mine, Sudbury, Ont., Canada BM 1922,145 Cobalt, Ont., Canada men M12176, where a 9% Ni variation is inversely proportional to a 2% Fe and 7% Co variation. The BM 1959,462 Cochabamba, Bolivia USNM 113044 Temagami, Ont., Canada specimens also show a wide range of substitution of BM 1434 Musen, Westphalia, Germany S by As, which falls within the solid solution limits USNM 120381 Art Ahmane, Bou Azzer, Morocco BM 1917,285 Sudbury, Ont., Canada of Yund (1962) at 4500 C. Non-metal chemical zon- ing is observed in specimen UNSW, where a 9% As UNSW Ferro, Slovakia, Czechoslovakia variation is inversely proportional to a 4% S varia- BM 1933,371 Farvic Mine, Gwanda, Rhodesia tion. 1060 BA YLlSS: GERSDORFFITE From each gersdorffite specimen, a sample was c1udes 24 reflections forbidden by space group Pa3, selected with an approximately equidimensional were collected. These integrated intensities were well-developed cube form {100}. The preliminary corrected for background and scaled so that the alignment along an a axis was made on a precession maximum observed relative intensity (l) is 1000. camera using unfiltered Mo radiation. These pre- Values of 1 are approximately equal to the limit of cession photographs hkO and hOI confirmed the detection at the 3a confidence level. presence of reflections 010 and 110 observed in the The crystallographic axes of each sample were powder diffractometer patterns. Simple twinning chosen so that 1120~ 1210,and if possible 1010> 1001 was not observed by either reflection splitting or the > 1100' This is satisfied by only one of the six presence of strong 210 and strong 120 reflections. possible orthogonal orientations. If 1010 = 1001 = Strong 210 and 120 reflections may result from a 1100,then three different orthogonal orientations are (110) twin plane as shown by iron cross twinning in possible.